Queue Realtime V2 response.create while active (#17306)

Builds on #17264.

- queues Realtime V2 `response.create` while an active response is open,
then flushes it after `response.done` or `response.cancelled`
- requests `response.create` after background agent final output and
steering acknowledgements
- adds app-server integration coverage for all `response.create` paths

Validation:
- `just fmt`
- `cargo check -p codex-app-server --tests`
- `git diff --check`
- CI green

---------

Co-authored-by: Codex <noreply@openai.com>

codex-rs/codex-api/src/endpoint/realtime_websocket/methods.rs

@@ -419,7 +419,9 @@ impl RealtimeWebsocketEvents {
             }
             RealtimeEvent::SessionUpdated { .. }
             | RealtimeEvent::AudioOut(_)
+            | RealtimeEvent::ResponseCreated(_)
             | RealtimeEvent::ResponseCancelled(_)
+            | RealtimeEvent::ResponseDone(_)
             | RealtimeEvent::ConversationItemAdded(_)
             | RealtimeEvent::ConversationItemDone { .. }
             | RealtimeEvent::Error(_) => {}
@@ -724,6 +726,8 @@ mod tests {
     use codex_protocol::protocol::RealtimeHandoffRequested;
     use codex_protocol::protocol::RealtimeInputAudioSpeechStarted;
     use codex_protocol::protocol::RealtimeResponseCancelled;
+    use codex_protocol::protocol::RealtimeResponseCreated;
+    use codex_protocol::protocol::RealtimeResponseDone;
     use codex_protocol::protocol::RealtimeVoice;
     use http::HeaderValue;
     use pretty_assertions::assert_eq;
@@ -999,7 +1003,9 @@ mod tests {
 
         assert_eq!(
             parse_realtime_event(payload.as_str(), RealtimeEventParser::RealtimeV2),
-            None
+            Some(RealtimeEvent::ResponseDone(RealtimeResponseDone {
+                response_id: None
+            }))
         );
     }
 
@@ -1013,10 +1019,9 @@ mod tests {
 
         assert_eq!(
             parse_realtime_event(payload.as_str(), RealtimeEventParser::RealtimeV2),
-            Some(RealtimeEvent::ConversationItemAdded(json!({
-                "type": "response.created",
-                "response": {"id": "resp_created_1"}
-            })))
+            Some(RealtimeEvent::ResponseCreated(RealtimeResponseCreated {
+                response_id: Some("resp_created_1".to_string())
+            }))
         );
     }
 

codex-rs/codex-api/src/endpoint/realtime_websocket/protocol_v2.rs

@@ -7,6 +7,8 @@ use codex_protocol::protocol::RealtimeEvent;
 use codex_protocol::protocol::RealtimeHandoffRequested;
 use codex_protocol::protocol::RealtimeInputAudioSpeechStarted;
 use codex_protocol::protocol::RealtimeResponseCancelled;
+use codex_protocol::protocol::RealtimeResponseCreated;
+use codex_protocol::protocol::RealtimeResponseDone;
 use serde_json::Map as JsonMap;
 use serde_json::Value;
 use tracing::debug;
@@ -47,23 +49,17 @@ pub(super) fn parse_realtime_event_v2(payload: &str) -> Option<RealtimeEvent> {
             .cloned()
             .map(RealtimeEvent::ConversationItemAdded),
         "conversation.item.done" => parse_conversation_item_done_event(&parsed),
-        "response.created" => Some(RealtimeEvent::ConversationItemAdded(parsed)),
+        "response.created" => Some(RealtimeEvent::ResponseCreated(RealtimeResponseCreated {
+            response_id: parse_response_event_response_id(&parsed),
+        })),
         "response.cancelled" => Some(RealtimeEvent::ResponseCancelled(
             RealtimeResponseCancelled {
-                response_id: parsed
-                    .get("response")
-                    .and_then(Value::as_object)
-                    .and_then(|response| response.get("id"))
-                    .and_then(Value::as_str)
-                    .map(str::to_string)
-                    .or_else(|| {
-                        parsed
-                            .get("response_id")
-                            .and_then(Value::as_str)
-                            .map(str::to_string)
-                    }),
+                response_id: parse_response_event_response_id(&parsed),
             },
         )),
+        "response.done" => Some(RealtimeEvent::ResponseDone(RealtimeResponseDone {
+            response_id: parse_response_event_response_id(&parsed),
+        })),
         "error" => parse_error_event(&parsed),
         _ => {
             debug!("received unsupported realtime v2 event type: {message_type}, data: {payload}");
@@ -72,6 +68,21 @@ pub(super) fn parse_realtime_event_v2(payload: &str) -> Option<RealtimeEvent> {
     }
 }
 
+fn parse_response_event_response_id(parsed: &Value) -> Option<String> {
+    parsed
+        .get("response")
+        .and_then(Value::as_object)
+        .and_then(|response| response.get("id"))
+        .and_then(Value::as_str)
+        .map(str::to_string)
+        .or_else(|| {
+            parsed
+                .get("response_id")
+                .and_then(Value::as_str)
+                .map(str::to_string)
+        })
+}
+
 fn parse_output_audio_delta_event(parsed: &Value) -> Option<RealtimeEvent> {
     let data = parsed
         .get("delta")